Search papers, labs, and topics across Lattice.
Tsinghua University
1
0
3
Effective keyframe extraction can boost MLLM performance by over 2% on complex video-guided tasks, revealing a critical link between video understanding and procedural learning.